CN115861986A - Non-standard product intelligent identification and loss prevention method based on supermarket self-service checkout system - Google Patents
Non-standard product intelligent identification and loss prevention method based on supermarket self-service checkout system Download PDFInfo
- Publication number
- CN115861986A CN115861986A CN202211651378.9A CN202211651378A CN115861986A CN 115861986 A CN115861986 A CN 115861986A CN 202211651378 A CN202211651378 A CN 202211651378A CN 115861986 A CN115861986 A CN 115861986A
- Authority
- CN
- China
- Prior art keywords
- product
- site
- calculating
- buyer
- standard
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Abstract
The invention discloses a non-standard product intelligent identification and loss prevention method based on a supermarket self-service checkout system. The type similarity of every two products is represented by judging whether the two sites are likely to be impacted, the complicated type feature comparison problem is simplified into a distance calculation problem, and the non-standard identification efficiency is higher. In addition, through the set g of intelligent recognition i Each type of product in the supermarket self-service checkout system is attached with an RFID label with a unique label, and when a buyer checks out the amount of the product, the supermarket self-service checkout system automatically scans RThe FID label obtains the unique label code, then the unique label code is compared with the commodity bar code of each commodity scanned when the buyer pays for the information consistency, the non-standard article damage prevention judgment is carried out according to the comparison result, and the problem that the non-standard article is missed to be scanned and paid is well solved.
Description
Technical Field
The invention relates to the technical field of commodity category identification and loss prevention, in particular to a non-standard product intelligent identification and loss prevention method based on a supermarket self-service checkout system.
Background
The standard products generally refer to products with similar or identical functions and similar appearances, and products without great difference from one product to another, such as paper diapers, paper towels, folders and other types of products are all standard products. Non-standard articles, generally refer to products with large differences in style or function, such as clothes, bags, and the like. The classification of the standard articles and the non-standard articles is generally difficult, the manual classification is easy to make mistakes, and when the quantity of the commodities needing to be classified is large, the defects of time consumption and complexity of the manual classification mode are more obvious.
In addition, compared with standard products (such as mobile phone films, tissues and the like), non-standard products (such as clothes, electric cookers and the like) are generally higher in value, and buyers generally bring greater loss if the non-standard products are subjected to missed scanning and payment compared with the standard products in a supermarket self-checkout system, so that in a scene of code scanning and payment by using the self-checkout system, how to prevent the buyers from bringing loss to the supermarket due to malicious missed scanning and payment of the non-standard products becomes a technical problem to be solved urgently.
Disclosure of Invention
The invention provides a non-standard product intelligent identification and loss prevention method based on a supermarket self-service checkout system, aiming at intelligently identifying non-standard products by a machine and carrying out loss prevention identification and treatment on malicious non-standard product missing scanning and missing payment behaviors of buyers.
In order to achieve the purpose, the invention adopts the following technical scheme:
the method for intelligently identifying the non-standard articles comprises the following steps:
s1, counting the offset of a site pixel of each site in a coordinate site image between every two adjacent frames in historical continuous n frames, and calculating the average offset speed of each site according to a frame rate;
s2, correspondingly marking the reference site i and each arbitrary site j in the coordinate site image asIs calculated as a vector composite velocity v ij The type of the product corresponding to the reference site i is a non-standard product;
s3, calculating the impact time T of the reference position i and the position j ij ;
S4, respectively calculating the pixel offset distances S of the impact meeting points of the reference point i and the point j i 、s j And respectively calculating the actual offset S of the pixel of the position point after the reference position point i and the position point j go through n continuous frames i 、S j ;
S5, judging whether | S is present i -S i Threshold th is less than or equal to |) 1 And | s j -S j Threshold th is less than or equal to |) 2 ,
If yes, judging the suspected same type of the products corresponding to the sites i and j respectively, and adding the site j into a set g corresponding to the reference site i i Then, go to step S6;
if not, judging that i and j are not the same, and adding the position j into the set g of the standard products j Performing the following steps;
s6, repeating the steps S2-S5 until all the sites in the coordinate site image are added into the set g i Or the set g j And determining the set g i The types of the products corresponding to all the sites in the standard are nonstandard articles.
Preferably, the lower left vertex of the rectangular coordinate site image is an origin of an xy-axis coordinate system, a horizontal axis coordinate value of each site in the coordinate site image is a first weighted average of a decision efficiency parameter, a price sensitivity parameter and a ranking sensitivity parameter of a product corresponding to the site purchased by different buyers during n consecutive frames, and a vertical axis coordinate value is a second weighted average of a store loyalty parameter and a product return rate parameter of the product corresponding to the site purchased by different buyers during n consecutive frames.
Preferably, the abscissa axis of each said locus in said coordinate locus image is located atScalar value x i The method is calculated by the following method steps:
a1, a supermarket self-service checkout system identifies a corresponding face image from a member face library through a member ID input by a buyer;
a2, taking a code scanning payment of the buyer on a product corresponding to the position point in the supermarket self-service checkout system as an instruction, acquiring a first frame face image of the buyer entering the position area and a last frame face image of the buyer exiting the position area from a monitoring database of the position area of the product corresponding to the position point, and then calculating an acquisition time difference between the last frame face image and the first frame face image as a staying time length in the position area representing the decision efficiency of the buyer for purchasing the product;
a3, obtaining a decision efficiency parameter value corresponding to the time interval in which the stay time falls by looking up a table
A4, calculating the mean value of decision efficiency parameters of all buyers for purchasing the same product corresponding to the site through the following formula (1)
In formula (1), k represents the kth buyer purchasing the same product corresponding to the location;
k represents a total number of buyers purchasing the product in a certain frame of the historical consecutive n frames;
a5, calculating the average of the historical prices of different buyers for purchasing the same product corresponding to the position through the following formula (2)Deviation of
Respectively representing the historical prices of the k +1 th buyer and the k-th buyer for purchasing the product, wherein the price data are sorted by k values;
a6, looking up a table to obtain the average deviation of the historical pricesThe price sensitivity parameter value corresponding to the falling deviation range->
A7, obtaining sales volume ranking of the products corresponding to the sites in the same products according to sales volume statistical data of the supermarket self-service checkout system, and then obtaining a ranking sensitivity parameter value v corresponding to the sales volume ranking by looking up a table rs ;
A8, calculating the first weighted average value as a horizontal axis coordinate value x of the coordinate point image of the point corresponding to the product by the following formula (3) i :
In the formula (3), a 1 、a 2 、a 3 Respectively representv rs In calculating x i The weight occupied by the time.
Preferably, each said location is in said coordinate location imageOrdinate value y of i The method is calculated by the following method steps:
b1, obtaining the times of the same product corresponding to the locus of each buyer k for the repurchase, and looking up a table to obtain a shop loyalty parameter value corresponding to the repurchase time range in which the repurchase times fall
In formula (4), L represents the total number of buyers who buy the same product again within one frame in n consecutive frames in the history;
k=1,2,…,L;
b3, calculating the return rate v of the product in a specified period according to the return data and the sales volume data of the supermarket self-service checkout system rr ;
B4, calculating the second weighted average value as a longitudinal axis coordinate value y of the position point corresponding to the product in the coordinate position point image through the following formula (5) i :
In the formula (5), b 1 、b 2 Respectively representv rr In calculating y i The weight occupied by the time.
Preferably, n =12 in n consecutive frames of the history indicates 12 months, that is, the coordinate location image of each frame is drawn based on the purchasing behavior data of the buyer in the current month;
the specified period of calculating the return rate of the product is one month.
Preferably, in step S3, the impact time T is calculated by the following formula (6) ij :
In formula (6), s represents a straight-line distance between the reference point i and the point j in the coordinate position point image of the first frame of the historical continuous n frames;
in step S4, S i Calculated by the following equation (7):
s j calculated by the following equation (8):
in step S4, S i The accumulated value of the actual offset of the position point pixel of the reference position point i between every two adjacent frames in the historical continuous n frames is obtained;
S j and the accumulated value of the actual offset of the pixel of the position point j between every two adjacent frames in the historical continuous n frames.
The invention also provides a non-standard article loss prevention method based on the supermarket self-service checkout system, which comprises the following steps:
l1, set g identified by the intelligent non-standard identification method of any one of claims 1 to 6 i Each type of product in (1) is a non-standard product to which an RFID tag having a unique tag code is attached;
l2, during checkout, automatically reading each RFID label in the shopping basket by an RFID reader-writer arranged in the supermarket self-checkout system;
l3, the buyer scans the codes of each purchased commodity through the supermarket self-service checkout system, after each code scanning, the supermarket self-service checkout system compares the scanned commodity bar code information with the read unique tag codes contained in the RFID tags in a consistent manner, and records the comparison result;
l4, the supermarket self-service checkout system takes the payment completion of the buyer as an instruction, judges whether the read unique tag codes contained in each RFID tag are compared with the information consistency of the corresponding commodity bar codes,
if so, judging that the loss prevention detection is passed;
if not, judging that the loss prevention detection is not passed and prompting and alarming.
Preferably, the set g is formed i The method comprises the following steps:
l11, counting the position pixel offset of each position in the coordinate position image between every two adjacent frames in historical continuous n frames, and calculating the average offset speed of each position according to the frame rate;
l12, respectively marking the corresponding reference position i and each arbitrary position j in the coordinate position image asIs calculated as a vector composite velocity v ij The type of the product corresponding to the reference site i is a non-standard product;
l13, calculating the collision time T of the reference point i and the point j ij ;
L4, respectively calculating the pixel offset distances s of the reference point i and the impact meeting point of the point j i 、s j And respectively calculating the actual offset S of the pixel of the position point after the reference position point i and the position point j go through n continuous frames i 、S j ;
L5, judging whether | s is present i Threshold th is less than or equal to-i |) 1 And | s j Threshold th is less than or equal to-j | 2 ,
If yes, judging the suspected same type of the products corresponding to the sites i and j respectively, and adding the site j into the set corresponding to the reference site ig i Then, the step L6 is carried out;
if not, judging that i and j are not the same, and adding the position j into the set g of the standard products j Performing the following steps;
l6, repeating the steps L2-L5 until all sites in the coordinate site image are added to the set g i Or the set g j And determining the set g i The types of products corresponding to all the sites in the product are nonstandard products.
The invention has the following beneficial effects:
1) And a semi-automatic non-standard identification scheme is adopted, so that the identification accuracy is higher. The type of the product corresponding to the reference point i is recognized and known as a non-standard product, and the standard product and the non-standard product can be quickly identified for a huge number of commodities by searching the similarity of the type characteristics of other products and the product corresponding to the reference point i.
2) The type similarity of every two products is represented by judging whether the two sites are likely to be impacted, the complicated type feature comparison problem is simplified into a distance calculation problem, and the non-standard identification efficiency is higher. The method comprises the steps of constructing each product as a position point in a coordinate position point image under an xy axis coordinate system, reflecting the change characteristics of characteristic data influencing the product type in different periods to the track change of the corresponding position point in the coordinate position point image, calculating whether each position point j in the coordinate position point image has the possibility of colliding with a reference position point i after experiencing historical continuous n frames, directly judging the position point j as being not the same as the reference position point i if the possibility does not exist, and if the possibility does exist, judging that the position point j satisfies | s |, if the possibility exists, judging that the position point j is not the same as the reference position point i, and if the position point j satisfies | s i Threshold th is less than or equal to-i |) 1 And | s j -j ≦ threshold th 2 If yes, j and i are judged to be similar, and if not, j and i are also judged to be dissimilar.
3) Set g by pair intelligent recognition i Attaching an RFID label with a unique label to each product of which the type is non-standard goods, scanning the RFID label by a supermarket self-service checkout system to obtain a unique label code when a buyer checks out the commodity, comparing the information consistency with the commodity bar code of each commodity scanned when the buyer pays, and if the unique label codes of all the RFID labels are uniqueAnd if the label codes are compared successfully, judging that the non-standard object loss prevention detection of the purchasing behavior passes, otherwise, judging that the loss prevention detection does not pass and performing prompt alarm, and well solving the problem of missing scanning and missing payment of the non-standard object.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required to be used in the embodiments of the present invention will be briefly described below. It is obvious that the drawings described below are only some embodiments of the invention, and that for a person skilled in the art, other drawings can be derived from them without inventive effort.
Fig. 1 is a diagram illustrating implementation steps of a non-standard intelligent identification method according to an embodiment of the present invention;
fig. 2 is a diagram illustrating implementation steps of a non-standard article damage prevention method based on a supermarket self-service checkout system according to an embodiment of the present invention;
fig. 3 is a schematic diagram of a coordinate site image in an xy-axis coordinate system.
Detailed Description
The technical scheme of the invention is further explained by the specific implementation mode in combination with the attached drawings.
Wherein the showings are for the purpose of illustration only and are shown by way of illustration only and not in actual form, and are not to be construed as limiting the present patent; for a better explanation of the embodiments of the present invention, some parts of the drawings may be omitted, enlarged or reduced, and do not represent the size of an actual product; it will be understood by those skilled in the art that certain well-known structures in the drawings and descriptions thereof may be omitted.
The same or similar reference numerals in the drawings of the embodiments of the present invention correspond to the same or similar components; in the description of the present invention, it should be understood that if the terms "upper", "lower", "left", "right", "inner", "outer", etc. are used for indicating the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, it is only for convenience of description and simplification of description, but it is not indicated or implied that the referred device or element must have a specific orientation, be constructed in a specific orientation and be operated, and therefore, the terms describing the positional relationship in the drawings are only used for illustrative purposes and are not to be construed as limitations of the present patent, and the specific meanings of the terms may be understood by those skilled in the art according to specific situations.
In the description of the present invention, unless otherwise explicitly specified or limited, the term "connected" or the like, if appearing to indicate a connection relationship between components, is to be understood broadly, for example, as being either fixedly connected, detachably connected, or integrated; can be mechanically or electrically connected; they may be directly connected or indirectly connected through intervening media, or may be connected through one or more other components or may be in an interactive relationship with one another. The specific meanings of the above terms in the present invention can be understood in specific cases to those skilled in the art.
The non-standard product intelligent identification method provided by the embodiment of the invention, as shown in figure 1, comprises the following steps:
s1, counting the offset of a site pixel of each site in a coordinate site image between every two adjacent frames in historical continuous n frames, and calculating the average offset speed of each site according to a frame rate;
as shown in fig. 3, the lower left vertex of the rectangular coordinate site image is the origin of the xy-axis coordinate system, the horizontal axis coordinate value of each site in the coordinate site image is the first weighted average of the decision efficiency parameter, the price sensitivity parameter, and the ranking sensitivity parameter of the product corresponding to the site purchased by different buyers during n consecutive frames, and the vertical axis coordinate value is the second weighted average of the store loyalty parameter and the product return rate parameter of the product corresponding to the site purchased by different buyers during n consecutive frames.
The continuous n frames are preferably continuous 12 frames, and the coordinate location image of each frame is drawn based on the purchasing behavior data of the buyer in the current month.
Horizontal axis coordinate value x of each position point in coordinate position point image i The method is calculated by the following method steps:
a1, a supermarket self-service checkout system identifies a corresponding face image from a member face library through a member ID input by a buyer;
a2, taking code scanning payment of a product corresponding to the point by the buyer in a supermarket self-service checkout system as an instruction, acquiring a first frame of face image of the product entering the point area and a last frame of face image of the product leaving the point area from a monitoring database of the point area, and calculating the acquisition time difference of the last frame of face image and the first frame of face image as the stay time of the product in the point area representing the decision efficiency of the buyer to purchase the product; the method for acquiring the first frame face image and the last frame face image of the buyer entering the corresponding category area is briefly as follows:
step A1, matching the face image of the buyer who is currently self-checkout from a member face library according to a member ID, wherein when the buyer scans a product, a self-checkout system can quickly acquire a category area where the product is displayed according to the category of the product (such as an electronic product) contained in scanned product information; and each class area is provided with one or a plurality of dedicated monitoring probes, the monitoring probes store all the acquired face data into a monitoring database corresponding to the class area, the system matches all the images with the same face in the face images identified in the step A1 in the monitoring database, and finally extracts the first frame face image and the last frame face image and calculates the acquisition time difference.
A3, looking up a table to obtain a decision efficiency parameter value corresponding to the time interval in which the stay time fallsFor example, the preset time intervals include 5 time intervals which are respectively 0-1 minute, 1-3 minutes, 3-5 minutes, 5-10 minutes and more than 10 minutes, wherein the time intervals of 0-1 minute, 1-3 minutes, 3-5 minutes, 5-10 minutes and more than 10 minutes respectively correspond to decision efficiency parameter values of 5, 4, 3, 2 and 1, if the stay time of a buyer in the category area is 30 seconds, the corresponding decision efficiency parameter value is 5, and the larger the decision efficiency parameter value is, the higher the decision efficiency of the buyer for purchasing the product in the category area is represented;
the main characteristic differences of the standard and non-standard are shown in the following table a:
efficiency of decision making | Price sensitivity | Product rank sensitivity | Shop loyalty | Rate of return | |
Standard article | Height of | High (a) | Height of | Is low in | Is low in |
Non-standard article | Is low in | Is low with | Is low in | Height of | Height of |
TABLE a
A4, calculating the mean value of decision efficiency parameters of all buyers for purchasing the same product corresponding to the site through the following formula (1)
In formula (1), k represents the kth buyer purchasing the same product corresponding to the site;
k represents the total number of buyers purchasing the product in a certain frame of the historical continuous n frames;
a5, calculating the average deviation of the historical prices of different buyers for purchasing the same product corresponding to the site by the following formula (2)
Respectively representing the historical prices of the k +1 th buyer and the k-th buyer for purchasing the product, wherein the price data are sorted by k values; for example, if buyer 1-5 purchases product A in a certain month, buyer 1 purchases product A with a price of 10 Yuan, buyer 2 purchases product A with a price of 8 Yuan, buyer 3 purchases product A with a price of 6 Yuan, buyer 4 purchases product A with a price of 14 Yuan, and buyer 5 purchases product A with a price of 12 Yuan, then based on formula (2), what is considered is that>According to Table a, the buyer's price sensitivity is typically higher for a standard than for a non-standard, i.e., </or >>The smaller in general;
a6, obtaining the average deviation of the historical prices by looking up a tableThe price sensitivity parameter value corresponding to the falling deviation range->Such as->Values which lie in a deviation range 0-5 correspond>A value of 5, corresponding to a deviation range of 5-10>A value of 4 which corresponds to a deviation range 10-15>A value of 3 corresponding to a deviation range 15-20>A value of 2 corresponding to a deviation range 20-25>Is 1;
a7, obtaining sales ranking of products corresponding to the point in the same kind of products according to sales statistical data of the supermarket self-service checkout system, and then obtaining a ranking sensitivity parameter value v corresponding to the sales ranking by looking up a table rs For example, the product corresponding to the site is a rice cooker with the model number 001, one of supermarkets has 5 types of rice cookers with the models of 001, 002, 003, 004 and 005, the sales figures in the month are respectively 002, 001, 003, 004 and 005, and the sales figures of the rice cookers are respectively ranked from the first value v to the fifth value v rs 5, 4, 3, 2, 1, respectively, the value v corresponding to the rice cooker of 001 rs =4;
A8, calculating a first weighted average value as a coordinate value x of a horizontal axis of a position point corresponding to the product in a coordinate position point image through the following formula (3) i :
In the formula (3), a 1 、a 2 、a 3 Respectively representv rs In calculating x i The weight occupied by the time.
The coordinate value y of the longitudinal axis of each position point in the coordinate position point image i The method is calculated by the following method steps:
b1, obtaining the times of the same product corresponding to each buyer k repurchase site, and looking up a table to obtain a shop loyalty parameter value corresponding to the repurchase time range in which the repurchase times fallFor example, buyer 1-3 buys product A again in the same month, wherein buyer 1 buys product A again once, buyer 2 buys product A again 4 times, buyer 3 buys product A again 6 times, the preset buys time range is assumed to be buys 1-3 times, buys 3-5 times and buys 5 times or more, then the buys time of buyer 1 falls within the buys 1-3 times range, the buys time of buyer 2 falls within the buys 3-5 times range, the buys time of buyer 3 falls within the buys 5 times range, and the predetermined buys time ranges are assumed to correspond to values of 1-3 times, 3-5 times and 5 times or more respectively>3, 2, 1, buyer 1 buys product A again with a value ≥>Buyer 2 value @>Buyer 3's value @>
In formula (4), L represents the total number of buyers purchasing the same product in one frame in n consecutive frames in the history;
k=1,2,…,L;
b3, calculating the return rate v of the product in a specified period (in one frame of historical continuous n frames) according to the return data and the sales data of the supermarket self-service checkout system rr ,v rr Dividing the product by the returned quantity and the sold quantity in the current month;
b4, calculating a second weighted average value as a longitudinal axis coordinate value y of the corresponding position point of the product in the coordinate position point image through the following formula (5) i :
In the formula (5), b 1 、b 2 Respectively representv rr In calculating y i The weight occupied by the time.
S2, respectively marking the corresponding reference position i and each arbitrary position j in the coordinate position image asIs calculated by calculating the vector composite velocity v ij The type of the product corresponding to the reference point i is a non-standard product,vector resultant velocity v ij For the conventional mathematical solution of the resultant velocity in the belt direction, it should be noted that for the convenience of v ij Is calculated by v ij Determining the offset direction of the time position points i and j as the direction of offsetting from the first frame of the history succession n to the second frame; />
S3, calculating the impact time T of the reference position point i and the reference position point j ij The calculation manner is expressed by the following formula (6):
in the formula (6), s represents the straight-line distance between a reference point i and a point j in the first frame coordinate position point image of the historical continuous n frames;
s4, respectively calculating the offset distances S of the pixels of the impact meeting points of the reference point i and the reference point j i 、s j And respectively calculating the actual offset S of the site pixel of the reference site i and the site j after the reference site i and the site j go through n continuous frames i 、S j ;
s i Calculated by the following equation (7):
s j calculated by the following equation (8):
S i the accumulated value of the actual offset of the position point pixel of the reference position point i between every two adjacent frames in the historical continuous n frames is obtained;
S j the accumulated value of the actual offset of the pixel of the position point j between every two adjacent frames in the historical continuous n frames is used as the position point.
S5, judging whether | S is present i -i ≦ threshold th 1 And | s j -j ≦ threshold th 2 ,
If yes, then determineProducts corresponding to the positions i and j are suspected to be the same type, and the position j is added into a set g corresponding to the reference position i i Then, the step S6 is carried out;
if not, judging that i and j are not the same, and adding the point j into the set g of the standard products j The preparation method comprises the following steps of (1) performing;
s6, repeating the steps S2-S5 until all the sites in the coordinate site image are added into the set g i Or set g j And determine set g i The types of products corresponding to all the sites in the product are nonstandard products.
In conclusion, the invention adopts a semi-automatic non-standard article identification scheme, and the identification accuracy is higher. The type of the product corresponding to the reference point i is known and acknowledged, and the invention can quickly identify the standard product and the non-standard product for a huge number of commodities by searching the similarity of the type characteristics of other products and the product corresponding to the reference point i. In addition, the type similarity of every two products is represented by judging whether the two sites are likely to be impacted, the complicated type characteristic comparison problem is simplified into a distance calculation problem, and the non-standard product identification efficiency is higher.
The invention also provides a non-standard article loss prevention method based on the supermarket self-service checkout system, as shown in fig. 2, the method comprises the following steps:
l1, set g identified by the intelligent non-standard article identification method i Each type of product in (1) is a non-standard product to which an RFID tag having a unique tag code is attached;
l2, during checkout, automatically reading each RFID label in the shopping basket by an RFID reader-writer arranged in the supermarket self-checkout system;
l3, scanning the code of each purchased commodity by the buyer through a supermarket self-service checkout system, after each code scanning, carrying out consistency comparison on the scanned commodity bar code information and the read unique tag codes contained in the RFID tags by the supermarket self-service checkout system, and recording a comparison result;
l4, the supermarket self-service checkout system takes the payment completion of the buyer as an instruction, judges whether the read unique tag codes contained in each RFID tag are compared with the information consistency of the corresponding commodity bar codes,
if so, judging that the loss prevention detection is passed;
if not, judging that the loss prevention detection is not passed and prompting and alarming.
In conclusion, the invention adopts the set g of intelligent identification i When a buyer pays off the goods, a supermarket self-service payment system autonomously scans the RFID tag to obtain a unique tag code, then compares the unique tag code with the commodity bar code of each piece of goods scanned by the buyer when paying, if the unique tag codes of all the RFID tags are compared successfully, the non-standard goods of the purchasing behavior are judged to be passed through the anti-damage detection, otherwise, the anti-damage detection is judged not to be passed, and prompt alarm is carried out, so that the problem that the non-standard goods are missed to be scanned and paid is well solved.
It is to be understood that the above-described embodiments are merely preferred embodiments of the invention and that the technical principles herein may be applied. It will be understood by those skilled in the art that various modifications, equivalents, changes, and the like can be made to the present invention. However, such variations are within the scope of the invention as long as they do not depart from the spirit of the invention. In addition, certain terminology used in the description and claims of the present application is not limiting, but is used for convenience only.
Claims (8)
1. A non-standard product intelligent identification method is characterized by comprising the following steps:
s1, counting the offset of a site pixel of each site in a coordinate site image between every two adjacent frames in historical continuous n frames, and calculating the average offset speed of each site according to a frame rate;
s2, correspondingly marking the reference site i and each arbitrary site j in the coordinate site image asIs calculated as a vector composite velocity v ij The type of the product corresponding to the reference site i is a non-standard product;
s3, calculating the impact time T of the reference position i and the position j ij ;
S4, respectively calculating the pixel offset distances S of the impact meeting points of the reference point i and the point j i 、s j And respectively calculating the actual offset S of the pixel of the position point after the reference position point i and the position point j go through n continuous frames i 、S j ;
S5, judging whether | S is present i -S i Threshold th is less than or equal to |) 1 And | s j -S j Threshold th is less than or equal to |) 2 ,
If yes, judging the suspected same type of the products corresponding to the sites i and j respectively, and adding the site j into a set g corresponding to the reference site i i Then, the step S6 is carried out;
if not, judging that i and j are not the same, and adding the position j into the set g of the standard products j Performing the following steps;
s6, repeating the steps S2-S5 until all the sites in the coordinate site image are added into the set g i Or the set g j And determining the set g i The types of products corresponding to all the sites in the product are nonstandard products.
2. The method of claim 1, wherein a lower left vertex of the coordinate site image of the rectangle is an origin of an xy-axis coordinate system, a horizontal axis coordinate value of each of the sites in the coordinate site image is a first weighted average of a decision efficiency parameter, a price sensitivity parameter, and a ranking sensitivity parameter of different buyers purchasing products corresponding to the site during n consecutive frames, and a vertical axis coordinate value is a second weighted average of a shop loyalty parameter and a product return rate parameter of different buyers purchasing products corresponding to the site during n consecutive frames.
3. The intelligent non-standard product identifying method as claimed in claim 2, wherein each site has x coordinate value of abscissa axis in the coordinate site image i By the following method stepsAnd calculating to obtain:
a1, a supermarket self-service checkout system identifies a corresponding face image from a member face library through a member ID input by a buyer;
a2, taking a code scanning payment of the buyer on a product corresponding to the position point in the supermarket self-service checkout system as an instruction, acquiring a first frame face image of the buyer entering the position area and a last frame face image of the buyer exiting the position area from a monitoring database of the position area of the product corresponding to the position point, and then calculating an acquisition time difference between the last frame face image and the first frame face image as a staying time length in the position area representing the decision efficiency of the buyer for purchasing the product;
a3, looking up a table to obtain a decision efficiency parameter value corresponding to the time interval in which the stay time falls
A4, calculating a mean value of decision efficiency parameters of all buyers for purchasing the same product corresponding to the site through the following formula (1)
In formula (1), k represents the kth buyer purchasing the same product corresponding to the location;
k represents the total number of buyers purchasing the product in a certain frame of the historical continuous n frames;
a5, calculating the average of the historical prices of different buyers for purchasing the same product corresponding to the position through the following formula (2)Deviation of
Respectively representing the historical prices of the k +1 th buyer and the k-th buyer for purchasing the product, wherein the price data are sorted by k values;
a6, obtaining the average deviation of the historical prices by looking up a tableThe price sensitivity parameter value corresponding to the falling deviation range->
A7, obtaining sales ranking of the products corresponding to the sites in the same products according to sales statistical data of the supermarket self-service checkout system, and then obtaining a ranking sensitivity parameter value v corresponding to the sales ranking by looking up a table rs ;
A8, calculating the first weighted average value as a horizontal axis coordinate value x of the position point corresponding to the product in the coordinate position point image through the following formula (3) i :
4. The intelligent non-standard object recognition method according to claim 2, wherein each of the sites has a longitudinal axis coordinate value y in the coordinate site image i The method is calculated by the following method steps:
b1, obtaining the times of repurchasing the same product corresponding to the locus by each buyer k, and looking up a table to obtain a shop loyalty parameter value corresponding to the scope of the repurchasing times of the repurchasing
In formula (4), L represents the total number of buyers who buy the same product again within one frame in n consecutive frames in the history;
k=1,2,…,L;
b3, calculating the return rate v of the product in a specified period according to the return data and the sales volume data of the supermarket self-service checkout system rr ;
B4, calculating the second weighted average value as a longitudinal axis coordinate value y of the position corresponding to the product in the coordinate position image through the following formula (5) i :
5. The intelligent non-standard object identification method according to claim 4, wherein n =12 in n consecutive frames in history represents 12 months, namely the coordinate location image of each frame is drawn based on the current month of buyer purchasing behavior data;
the specified period for calculating the return rate of the product is one month.
6. The intelligent non-standard object recognition method according to claim 1, wherein in step S3, the impact time T is calculated by the following formula (6) ij :
In formula (6), s represents a straight-line distance between the reference point i and the point j in the coordinate position point image of the first frame of historical continuous n frames;
in step S4, S i Calculated by the following equation (7):
s j calculated by the following equation (8):
in step S4, S i The accumulated value of the actual offset of the position point pixel of the reference position point i between every two adjacent frames in the historical continuous n frames is obtained;
S j and the accumulated value of the actual offset of the pixel of the position point j between every two adjacent frames in the historical continuous n frames.
7. A non-standard article loss prevention method based on a supermarket self-service checkout system is characterized by comprising the following steps:
l1, set g identified by the intelligent non-standard identification method of any one of claims 1 to 6 i Each type of product in (1) is a non-standard product to which an RFID tag having a unique tag code is attached;
l2, during checkout, automatically reading each RFID label in the shopping basket by an RFID reader-writer arranged in the supermarket self-checkout system;
l3, the buyer scans the codes of each purchased commodity through the supermarket self-service checkout system, after each code scanning, the supermarket self-service checkout system compares the scanned commodity bar code information with the read unique tag codes contained in the RFID tags in a consistent manner, and records the comparison result;
l4, the supermarket self-service checkout system takes the payment completion of the buyer as an instruction, judges whether the read unique tag codes contained in each RFID tag are compared with the information consistency of the corresponding commodity bar codes,
if so, judging that the loss prevention detection is passed;
if not, judging that the loss prevention detection is not passed and prompting and alarming.
8. Non-standard loss prevention method based on supermarket self-checkout system according to claim 7, characterized in that said set g is formed i The method comprises the following steps:
l11, counting the position pixel offset of each position in the coordinate position image between every two adjacent frames in historical continuous n frames, and calculating the average offset speed of each position according to the frame rate;
l12, respectively marking the corresponding reference position i and each arbitrary position j in the coordinate position image asSaid average offset velocity calculation vector composite velocity v ij The type of the product corresponding to the reference site i is nonMarking;
l13, calculating the impact time T of the reference position i and the position j ij ;
L4, respectively calculating the pixel offset distances s of the reference point i and the impact meeting point of the point j i 、s j And respectively calculating the actual offset S of the pixel of the position point after the reference position point i and the position point j go through n continuous frames i 、S j ;
L5, judging whether | s is i -S i Threshold th is less than or equal to |) 1 And | s j -S j Threshold th is less than or equal to |) 2 ,
If yes, judging the suspected same type of the products corresponding to the sites i and j respectively, and adding the site j into a set g corresponding to the reference site i i Then, the step L6 is carried out;
if not, judging that i and j are not of the same type, and adding the site j into a set g of the standard articles j The preparation method comprises the following steps of (1) performing;
l6, repeating the steps L2-L5 until all sites in the coordinate site image are added to the set g i Or the set g j And determining the set g i The types of products corresponding to all the sites in the product are nonstandard products.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211651378.9A CN115861986B (en) | 2022-12-21 | 2022-12-21 | Non-standard intelligent identification and loss prevention method based on supermarket self-service checkout system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211651378.9A CN115861986B (en) | 2022-12-21 | 2022-12-21 | Non-standard intelligent identification and loss prevention method based on supermarket self-service checkout system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115861986A true CN115861986A (en) | 2023-03-28 |
CN115861986B CN115861986B (en) | 2023-07-28 |
Family
ID=85653636
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211651378.9A Active CN115861986B (en) | 2022-12-21 | 2022-12-21 | Non-standard intelligent identification and loss prevention method based on supermarket self-service checkout system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115861986B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117132845A (en) * | 2023-10-27 | 2023-11-28 | 浙江由由科技有限公司 | Snack classification method by scanning code coordinated image recognition and checking through partition bars |
CN117152540A (en) * | 2023-10-27 | 2023-12-01 | 浙江由由科技有限公司 | Intelligent pricing method for fresh goods considering display position, sales volume and classification precision |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH11175597A (en) * | 1997-12-10 | 1999-07-02 | Hitachi Ltd | Merchandise selection behavior information calculating method and its execution system |
CN111260005A (en) * | 2020-01-16 | 2020-06-09 | 青岛创捷中云科技有限公司 | Self-service cash-receiving AI loss prevention system and method based on image recognition |
CN111666927A (en) * | 2020-07-08 | 2020-09-15 | 广州织点智能科技有限公司 | Commodity identification method and device, intelligent container and readable storage medium |
CN112308020A (en) * | 2020-11-20 | 2021-02-02 | 易启看(杭州)科技有限公司 | Off-cabinet state tracking method, device and system for non-standard commodities |
CN114820091A (en) * | 2021-01-27 | 2022-07-29 | 杭州天识信息科技有限公司 | Intelligent non-standard product classification system |
-
2022
- 2022-12-21 CN CN202211651378.9A patent/CN115861986B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH11175597A (en) * | 1997-12-10 | 1999-07-02 | Hitachi Ltd | Merchandise selection behavior information calculating method and its execution system |
CN111260005A (en) * | 2020-01-16 | 2020-06-09 | 青岛创捷中云科技有限公司 | Self-service cash-receiving AI loss prevention system and method based on image recognition |
CN111666927A (en) * | 2020-07-08 | 2020-09-15 | 广州织点智能科技有限公司 | Commodity identification method and device, intelligent container and readable storage medium |
CN112308020A (en) * | 2020-11-20 | 2021-02-02 | 易启看(杭州)科技有限公司 | Off-cabinet state tracking method, device and system for non-standard commodities |
CN114820091A (en) * | 2021-01-27 | 2022-07-29 | 杭州天识信息科技有限公司 | Intelligent non-standard product classification system |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117132845A (en) * | 2023-10-27 | 2023-11-28 | 浙江由由科技有限公司 | Snack classification method by scanning code coordinated image recognition and checking through partition bars |
CN117152540A (en) * | 2023-10-27 | 2023-12-01 | 浙江由由科技有限公司 | Intelligent pricing method for fresh goods considering display position, sales volume and classification precision |
CN117132845B (en) * | 2023-10-27 | 2024-01-05 | 浙江由由科技有限公司 | Snack classification method by scanning code coordinated image recognition and checking through partition bars |
CN117152540B (en) * | 2023-10-27 | 2024-01-09 | 浙江由由科技有限公司 | Intelligent pricing method for fresh goods considering display position, sales volume and classification precision |
Also Published As
Publication number | Publication date |
---|---|
CN115861986B (en) | 2023-07-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109214806B (en) | Self-service settlement method, device and storage medium | |
CN108846621B (en) | Inventory management system based on strategy module | |
CN115861986B (en) | Non-standard intelligent identification and loss prevention method based on supermarket self-service checkout system | |
EP3761281A1 (en) | Information processing system | |
CN212229682U (en) | Open type self-service sale system based on buyer position positioning | |
US8146811B2 (en) | Cart inspection for suspicious items | |
US8286868B2 (en) | Checkout methods and apparatus | |
WO2019019291A1 (en) | Settlement method and device of image recognition technology based on convolutional neural network | |
US20040078260A1 (en) | System for sales optimization utilizing biometric customer recognition technique | |
CN112785354B (en) | Commodity recommendation system for retail management platform | |
CN110033298A (en) | Information processing equipment and its control method, system and storage medium | |
CN109920178B (en) | Unmanned supermarket system | |
US10372998B2 (en) | Object recognition for bottom of basket detection | |
CN110622173A (en) | Detection of mislabeled products | |
CN114119001B (en) | Intelligent management and analysis platform for cash collection of commodity sales based on artificial intelligence | |
Bobbit et al. | Visual item verification for fraud prevention in retail self-checkout | |
JP7384516B2 (en) | Information processing device, product recommendation method, program, and system | |
CN113239793A (en) | Loss prevention method and device | |
JP3837475B2 (en) | Automated shopping system | |
CN109034887B (en) | Method, device and system for adjusting price of article | |
CN116611796A (en) | Exception detection method and device for store transaction data | |
CN112154488A (en) | Information processing apparatus, control method, and program | |
CN111507702A (en) | Self-service shopping method for unmanned supermarket, computer readable storage medium and system | |
CN112085538A (en) | Method and system for accelerating analysis of commodities based on big data | |
CN113723251A (en) | Loss prevention method and device |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |